Hey,
I've been hunting an issue in 2.6.14-rt16 (also present in
2.6.16-rt17), where the cpu rlimits were not being enforced.
I tracked it down in update_process_times, where we have the following:
#ifndef CONFIG_PREEMPT_RT
run_posix_cpu_timers(p);
#endif
Yea, that would do it. :)
Hmm.. Looking through the archives, I came up w/ the discussion here:
http://www.uwsg.iu.edu/hypermail/linux/kernel/0509.3/1579.html
I was just curious what folks are thinking about for a fix here? Should
we just create a soft timer that calls run_posix_cpu_timers?
thanks
-john